REGISTER | LOGIN
Breaking News
News & Analysis

ReRAM Goes 3D

6/30/2017 00:31 AM EDT
14 comments
NO RATINGS
More Related Links
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 2   >   >>
R_Colin_Johnson
User Rank
Author
Re: Selector mechanism?
R_Colin_Johnson   7/5/2017 8:59:36 AM
NO RATINGS
Here is a review all the current 3D-ReRAM approaches, as of 2016, which should answer most of your questions:

"3D resistive RAM cell design for high-density storage class memory—a review"

realjjj
User Rank
CEO
Re: Selector mechanism?
realjjj   7/5/2017 2:14:36 AM
NO RATINGS
What's your thionking? They do see the need for 64 layers for half the costs of horizontal so it would need to handle high aspect ratio better than the alternative. Assumed alternative would be ALD and doping?

R_Colin_Johnson
User Rank
Author
Re: Selector mechanism?
R_Colin_Johnson   7/4/2017 8:30:41 PM
NO RATINGS
Your the Man resistion. For whom do you work?

resistion
User Rank
Author
Re: Selector mechanism?
resistion   7/4/2017 8:12:37 PM
NO RATINGS
3D NAND or other sidewall-based memory has this problem or feature.

Also, how well does the plasma process behave at the bottom?

R_Colin_Johnson
User Rank
Author
Re: Selector mechanism?
R_Colin_Johnson   7/4/2017 7:58:57 PM
NO RATINGS
Vertical 3D ReRAM sidewall storage capacitance of the overhanging electrodes requires a long charging time and/or a large voltage--bingo. An engineering challenge worth five years effort? Thanks for hitting the nail on the head!

resistion
User Rank
Author
Re: Selector mechanism?
resistion   7/4/2017 5:44:19 PM
NO RATINGS
Generally, some self-nonlinearity is expected to be utilized for the selector function, so that actually having one may not be necessary, but whether it is sufficient depends on many constraints and conditions, including the architecture.

The problem with a 3D sidewall storage memory is the capacitance of the overhanging electrodes or the horizontal sheets, it either requires a long charging time and/or a large voltage. But 3D NAND has never been concerned about being fast.

resistion
User Rank
Author
Re: Do these guys have any idea
resistion   7/4/2017 5:41:27 PM
NO RATINGS
No, to my knowledge, Samsung, ITRI, Stanford, NTHU did this as well, with ALD or PEALD.

Shinobee
User Rank
Manager
Selector mechanism?
Shinobee   7/4/2017 5:38:55 PM
NO RATINGS
Using some sort of ALD to create a "3D VNAND-like" ReRAM will definitely have a higher throughput and better yield compared to the criss cross wire structure 3DXpoint, but the main challenge of 3D ReRAM that needs addressing now (to my knowledge) is the integration a selector mechanism, in this much tighter structure.  Does one exist for ReRAM in these types of structures?

R_Colin_Johnson
User Rank
Author
Re: Do these guys have any idea
R_Colin_Johnson   7/4/2017 3:06:43 PM
NO RATINGS
Here is Russia's explanation about what is "first" about their 3-D ReRAMs. In a nutshell they aim to stack 3-D ReRAMS vertically rather than horizontally using PEALD: "It seems that your readers are wondering what is so interesting about our research, given that manufacturing companies have already announced their own 3-D designs. The thing is, all the above-mentioned companies announced 3D Xpoint ReRAM and 3D cross-bar ReRAM (the so-called 3D horizontal-ReRAM) where the memory cell is horizontally oriented (horizontally stacked ReRams). A distinctive feature of this approach is the simplicity of the manufacturing process: Memory cells are just piled on top of each other. Hence, no development of new technological processes is needed, which is why this approach has already been tested and demonstrated by many companies. In our article, we are talking about 3D-Vertical ReRam architecture where memory cells are stacked in a vertical direction. It is believed that with 64 layers, the overall cost-per-bit is half that of H-ReRAM. However, manufacture of this type of architecture requires new technological processes to be developed, and there are major challenges researchers have been facing with regards to the development process. As far as I can tell, the results achieved in this area so far are rather modest (attempts were made by Samsung, Imec, Seoul Nat’l Univ, etc.). One of the key technological challenges is depositing a layer of a dielectric material containing oxygen vacancies on the vertical walls of the structure with an accuracy of pre-determining the thickness of the layer of several Angstroms. This requirement can be fulfilled only by ALD. However, even for ALD that is a very difficult task, as the method allows formation of films whose quality is too high, so it becomes really difficult to create oxygen vacancies in them. In our study, we have shown that PEALD enables us to both form oxygen vacancies and control their concentration. Although so far we have only demonstrated a “horizontal” cell (because it is easier to make), we believe our approach will help solve the core issue of 3D Vertical ReRAM and bring the technology to a qualitatively new level," said MIPT scientist Konstantin Egorov's in an exclusive EE Times interview.

R_Colin_Johnson
User Rank
Author
Re: Do these guys have any idea
R_Colin_Johnson   7/1/2017 6:44:36 AM
NO RATINGS
Yes, as early as 2016 Samsung, Western Digital, Crossbar and others also report 3-D ReRAM as the best candidate for next-gen universal memory. I'm querying the Russia's now about what is the "first" part of their claim, but I suspect it is that their peer reviewed journal took over a year to get published, whereas a press report only takes 1-day. Thanks for the tip, you are right.

Page 1 / 2   >   >>
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed